overviewofenergy’ exascaleearthsystem model(e3sm)’...overviewofenergy’ exascaleearthsystem...

22
Overview of Energy Exascale Earth System Model (E3SM) a new model from US Department of Energy from Ruby Leung, E3SM Chief Scientist E3SM (formerly ACME) started with CESM v1 as E3SM v0 E3SM v1 model components: Atmosphere (EAM): A fork of CAM5.3; increased verCcal levels from 30 to 72; CLUBB + MG2; MAM4 aerosols; other minor improvements Ocean (MPASOcean): A free surface ocean model with unstructured grid for variable resoluCon modeling Sea ice (MPASSea Ice): An unstructured grid model with CICE physics Land ice (MPASLand Ice): An unstructured grid model with iceocean interacCons beneath AntarcCc ice sheet Land (ELM): A fork of CLM4.5; new soil hydrology; CNP cycle and other new biogeochemistry representaCons River (MOSART): A scale adapCve river transport model

Upload: others

Post on 28-Feb-2021

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

Overview  of  Energy  Exascale  Earth  System  Model  (E3SM)  a new model from US Department of Energy from Ruby Leung, E3SM Chief Scientist  •  E3SM  (formerly  ACME)  started  with  CESM  v1  as  E3SM  v0  •  E3SM  v1  model  components:  

–  Atmosphere  (EAM):  A  fork  of  CAM5.3;  increased  verCcal  levels  from  30  to  72;  CLUBB  +  MG2;  MAM4  aerosols;  other  minor  improvements  

–  Ocean  (MPAS-­‐Ocean):  A  free  surface  ocean  model  with  unstructured  grid  for  variable  resoluCon  modeling  

–  Sea  ice  (MPAS-­‐Sea  Ice):  An  unstructured  grid  model  with  CICE  physics  –  Land  ice  (MPAS-­‐Land  Ice):  An  unstructured  grid  model  with  ice-­‐ocean  

interacCons  beneath  AntarcCc  ice  sheet  –  Land  (ELM):  A  fork  of  CLM4.5;  new  soil  hydrology;  CNP  cycle  and  other  

new  biogeochemistry  representaCons  –  River  (MOSART):  A  scale  adapCve  river  transport  model    

Page 2: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

E3SM  v1  model  configuraGons  •  Three  model  configuraCons:  

–  Low  resoluCon:  ~100  km  (ne30)  atmosphere/land;  60-­‐30  km  ocean/ice  

–  High  resoluCon:  ~25  km  (ne120)  atmosphere/land;  18-­‐6  km  ocean/ice  

–  Low  resoluCon  BGC:  low  resoluCon  model  +  land  and  ocean  BGC  

•  Model  release:  –  E3SM  v1  will  iniCally  be  released  with  (1)  a  low  resoluCon  model  configuraCon  and  DECK  simulaCons  and  (2)  a  high  resoluCon  model  configuraCon  with  a  1950s  simulaCon  –  tentaCvely  April  2018  

–  E3SM  v1  low  resoluCon  BGC  will  be  released  a]er  compleCon  of  low  resoluCon  BGC  experiments  (TBD)  

Page 3: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

Planned  simulaGons  •  CMIP6  DECK  experiments  using  low  resoluCon  configuraCon  for  

~  3  ensemble  members  –  tentaCvely  completed  by  March  2018  •  Water  cycle  experiments:  

–  Full  transient  forcing  simulaCons  at  low  and  high  resoluCon  for  1950-­‐2050  (similar  to  HighResMIP)  

–  Single  transient  forcing  simulaCons  at  low  and  high  resoluCon  for  1950-­‐2050  

•  BGC  experiments  (esCmate  carbon-­‐climate  and  carbon-­‐CO2  feedbacks):  –  (1)  Preindustrial  control  simulaCon  –  1850  for  250  years  –  (2)  Transient  forcing  1850-­‐2100  with  1850  GHG  concentraCons  –  Same  as  (2),  except  including  BGC  influence  of  transient  atmospheric  

CO2  concentraCon  –  Same  as  (2),  except  including  full  radiaCve  effects  of  GHG  

Page 4: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

Community  Earth  System  Model  (CESM)  for  CMIP6  

Office  of  Science  

Page 5: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

Target  CESM2  and  CMIP6  versions  (2018-­‐2019):      DECK  and  most  MIP  simulaGons:  

1.  AOGCM  physical  climate  (atmos  1o,  ocean  1o  ,  low-­‐top)  with  biogeochemistry  (CO2  emission  and/or  concentraCon  driven)  

2.  +  atmospheric  chemistry  (atmos  1o,  ocean  1o  ,  high-­‐top;  WACCM)    DECK  +  ScenarioMIP  Ger  1  simulaGons:  1.  AOGCM  physical  climate  (atmos  1/4o,  ocean  1o  ,  low-­‐top)  (currently  with  CESM1  at  this  resoluCon:    Cme  slice,  PI  control  100  years,  20th  century;    planning  on  RCP8.5  and  RCP2.6  in  2017)          2019-­‐2020:    simulaCons  with  AOGCM  atmos  1/4o  ,  ocean  1/10o  :    50  year  PI  control;  20C  and  RCP8.5      (currently  with  CESM1  at  this  resoluCon  have  50  year  perpetual  2000  and  RCP8.5  to  2050)  

Low-­‐top:  32  levels  up  to  40  km  High-­‐top:  72  levels  up  to  150  km  

Page 6: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

CESM2  parCcipaCon  in  CMIP6  

OMIP:    ocean  -­‐  sea-­‐ice  simulaCon  at  1/10  degree  resoluCon  forced  with  the  JRA-­‐55  atmospheric  data  sets  (in  addiCon  to  1  deg  ocean)  

MIPacronym MIPname Nameofprimarysponsor(s)AerChemMIP AerosolsandChemistryModelIntercomparisonProject Lamarque/Emmons/Liu(Wyoming)C4MIP CoupledClimateCarbonCycleModelIntercomparisonProject LindsayCFMIP CloudFeedbackModelIntercomparisonProject Medeiros/Kay(CU)/Klein(LLNL)DAMIP DetectionandAttributionModelIntercomparisonProject Tebaldi/ArblasterDCPP DecadalClimatePredictionProject Danabasoglu/MeehlGeoMIP GeoengineeringModelIntercomparisonProject Tilmes/MillsGMMIP GlobalMonsoonsModelIntercomparisonProject Fasullo/Kinter(COLA)HighResMIP HighResolutionModelIntercomparisonProject Neale/BacmeisterISMIP6 IceSheetModelIntercomparisonProjectforCMIP6 Lipscomb(LANL)/Otto-BliesnerLS3MIP LandSurface,SnowandSoilMoisture D.LawrenceLUMIP Land-UseModelIntercomparisonProject D.Lawrence/P.LawrenceOMIP/OCMIP OceanModelIntercomparisonProject Danabasoglu/LindsayPMIP PalaeoclimateModellingIntercomparisonProject Otto-BliesnerRFMIP RadiativeForcingModelIntercomparisonProject Gettelman/NealeScenarioMIP ScenarioModelIntercomparisonProject Meehl/O'Neill/P.LawrenceVolMIP VolcanicForcingsModelIntercomparisonProject Mills/Otto-BliesnerDataonlyCORDEX CoordinatedRegionalClimateDownscalingExperiment Mearns/GutowskiDynVar DynamicsandVariabilityoftheStratosphere-TroposphereSystem MarshSIMIP Sea-IceModelIntercomparisonProject Bailey/Holland/Jahn(CU)/Hunke(LANL)VIAAB VIAAdvisoryBoardforCMIP6 Mearns/O'NeillNotparticipatingFAFMIP Flux-Anomaly-ForcedModelIntercomparisonProjectNonlinMIP NonlinearclimateresponsestoCO2

Aixue  Hu  TentaCve  subject  to  resources  

Page 7: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

Improved  performance  for  CESM2  on  current  and  upcoming  supercomputers  has  been  a  focus  over  the  last  year  Throughput  esCmates  for  CESM2:    1.    1  deg  atmos  1  deg  ocean  standard  CESM2:      ~20  model  years  per  calendar  day  on  NCAR  Yellowstone    (climate  sensiCvity  ~4.5°)    2.  1/4  deg  atmos  1  deg  ocean  CESM2:    ~2.4  model  years  per  calendar  day  (on  Blue  Waters  and  Mira)    3.  high  top  1  deg  WACCM,  72  layers      ~6  model  years  per  day  on  NCAR  Yellowstone.    4.  1/4  deg  atmos  1/10  deg  ocean  CESM2:    ~1  model  year  per  day  (on  Blue  Waters)    

Page 8: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

CESM2  Global  surface  temperature    

(anomaly  w.r.t.1850-­‐1880)    

PrognosCc  aerosols  Same  model  (CESM2)  run  with  CMIP5  SO2  emissions  and  CMIP6  SO2  emissions  

Black:  HadCRUT4  Red:  #190:  CMIP6  Blue:  #192:  CMIP5  

Current  status  of  CESM2:  issue  when  switching  CMIP5  -­‐>  6  emissions  

Page 9: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

CESM2  Red:    CMIP6  Blue:    CMIP5  5  year  running  mean  1961-­‐1980  base  period  

HadCRUTv6.0  forcing  (10member)v6.1  forcing  (10member)

MIROC6  

Page 10: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

Differences  in  emissions  

10  

Red:  CMIP6  Blue:  CMIP5  

CMIP6  CMIP5   30%  

increase  

Same  SO2  emissions  in  1905  as  in  2010  

Page 11: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

Michou,  M.,  

Page 12: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

Differences  in  forcing  CMIP5  -­‐>  CMIP6    (example  for  decade  1970-­‐1980)  

12  

192:  CMIP5  

190:  CMIP6  

190-­‐192  

Annual  SWCF  (W/m2)   Annual  Low  Clouds  (%)  

Slide  from  J.  Bacmeister  

Page 13: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

NCAR  CMIP6  Planning  

January  2017  

January  2018  

January  2019  

January  2020  

January    2021  

NCAR  Yellowstone  

NCAR  Cheyenne    

CESM2  Validation  

CMIP6  SimulaGon  CMIP6  Data  to  ESGF  

CMIP6  Analysis  and  Papers  (CMIP  analysis  pla\orm)  

Ramp-­‐up  

2018  ProducGon  

2019  ProducGon  

CMIP6  Post-­‐Processing  

Ramp-­‐down  

CMIP6  SimulaGons  with  CESM2  

CESM2      1°  Release  

DOE  (NERSC,  Argonne);  Univ.  Illinois  (Blue  Waters)  

now  CESM2      1/4°  

Release  

2020  

1/4°  CESM1  simulaGons  

Page 14: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

END  

Page 15: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

Differences  in  emissions  

15  

Red:  CMIP6  Blue:  CMIP5  

Page 16: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

CMIP6  DECK  +  Tier  1  Requirements  

•  CESM2  1o  coupled/uncoupled  versions  – CESM2-­‐BGC:  ≈18,000  – CESM2  WACCM-­‐BGC:  ≈  5,500  – Total  cost:  ≈  240M  core-­‐hours  

•  CESM2  1/4o  coupled/uncoupled  version  – CESM2    1/4°:  ≈  400  years  – Total  cost:  ≈  50M  core-­‐hours  

Yellowstone core-hours; 1 year ≈ 700M core-hours

Page 17: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

CESM2  Red:    CMIP6  Blue:    CMIP5  5  year  running  mean  1961-­‐1980  base  period  

Page 18: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

Model  improvements  (1):  examples  from  MIROC6  (Update  from  MIROC5)  

AGCM  (T85L81)

 ・Shallow  convection

 ・High-‐‑‒Top  TOA  (3hPa  →  0.004  hPa)  ・SOA,  Oceanic  organic  Aerosol  ・Scattering  by  non-‐‑‒spherical  cloud  ice  ・Non-‐‑‒orographic  GWD  ・modified  CMT,  water  leak  fixed,  etc.

OGCM

 ・Higher  resolutions  (1.4°L50  →  1°L63)

 ・Tripole  coordinate  ・Improved  TKE  estimate  under  sea-‐‑‒ice

Land  Surface  Model

 ・Subgrid  snow  cover  distribution

・Wet  land  due  to  snow  melting

Current  status  of  CMIP6  experiments  (Done,  On-‐‑‒going,  Preparing)              DECK        CMIP6  historical  simulations  using  v6.0/v6.1  forcing  datasets        FAFMIP        DAMIP        OMIP        CFMIP        DCPP        HighresMIP        GMMIP        LS3MIP        RFMIP        ScenarioMIP        

Global-‐‑‒mean  SAT  (5yr  running-‐‑‒mean;  Ref:  1961/90)

HadCRUTv6.0  forcing  (10member)v6.1  forcing  (10member)

18  

Page 19: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

Differences  in  emissions  

19  

Red:  CMIP6  Blue:  CMIP5  

CMIP6  CMIP5  

Page 20: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

Transient  volcanic  forcing  in  CESM2  driven  by  emissions:  comparison  to  CMIP6  

20  

Page 21: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

Current  status  of  CESM2:  issue  when  switching  CMIP5  -­‐>  6  emissions  

Black:  HadCRUT4  Red:  #190:  CMIP6  Blue:  #192:  CMIP5  

Global  surface  temperature    (anomaly  w.r.t.1850-­‐1880)  

Same  model  version  used  in  both.  Only  differ  in  anthropogenic/bb  emissions  and  conCnuous  volcanoes  

Currently  working  on  understanding  reason  for  differences  and  remedies:  focus  on  cloud-­‐aerosol  interacCons  and  background  aerosols  

Page 22: OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’...OverviewofEnergy’ ExascaleEarthSystem Model(E3SM)’ a new model from US Department of Energy from Ruby Leung, E3SM Chief

CMIP6:  revised  Cmeline  and  workflow  

WACCM  (fixed  SSTs,  20  years)  

CESM2  (coupled,  100  years)  

Ocean  only  (2500  years)  

Land  only  (800  years)  

 CESM2  coupled  1850  control  (500  years)  

WACCM  coupled  1850  control  (150  years)  

22  

WACCM  (fixed  SSTs)  

CESM2  (coupled)  

CESM2:  20+  sypd  (1o,  32L)  WACCM:  6+  sypd  (1o,70L,  228  sp.)  Ocean  only:  100  sypd  Land  only:  150  sypd  

 CESM2  coupled  1850  control  (200  years)  

Start  of  DECK  &  20th  

Start  of  DECK  &  20th  

BGC  spinup  

Approximately  2  months  before  simulaCons  other  than  PI  control  can  get  started  

Days  a]er  model  is  frozen